1

Geomorphological records of extreme floods and their relationship to decadal-scale 1

climate change 2

3

S.A. Foulds*, H.M. Griffiths, M.G. Macklin, P.A. Brewer PA 4

5

Centre for Catchment and Coastal Research and River Basin Dynamics and 6

Hydrology Research Group, Department of Geography and Earth Sciences, 7

Aberystwyth University, Ceredigion, SY23 3DB 8

9

10

*Tel. +44 (0)1970 622606; Fax; +44 (0)1970 622659; E-mail. safoulds03@gmail.com 11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

2

Abstract 28

29

Extreme rainfall and flood events in steep upland catchments leave 30

geomorphological traces of their occurrence in the form of boulder berms, debris 31

cones, and alluvial fans. Constraining the age of these features is critical to 32

understanding (i) landscape evolution in response to past, present, and future 33

climate change; and (ii) the magnitude–frequency of extreme, ungauged floods in 34

small upland catchments. This research focuses on the Cambrian Mountains of 35

Wales, UK, where lichenometric dating of geomorphological features and 36

palaeohydrological reconstructions is combined with climatological data and 37

documentary flood records. Our new data from Wales highlight a distinct flood-rich 38

period between 1900 and 1960, similar to many other UK lichen-dated records. 39

However, this study sheds new light on the underlying climatic controls on upland 40

flooding in small catchments. Although floods can occur in any season, their timing 41

is best explained by the Summer North Atlantic Oscillation (SNAO) and shifts 42

between negative (wetter than average conditions with regular cyclonic flow and 43

flooding) and positive phases (drier than average conditions with less frequent 44

cyclonic flow and flooding), which vary from individual summers to decadal and 45

multidecadal periods. Recent wet summer weather, flooding, and boulder-berm 46

deposition in the UK (2007-2012) is related to a pronounced negative phase shift of 47

the SNAO. There is also increasing evidence that recent summer weather extremes 48

in the mid-latitudes may be related to Arctic amplification and rapid sea ice loss. If 49

this is the case, continuing and future climate change is likely to mean that (i) 50

unusual weather patterns become more frequent; and (ii) upland UK catchments will 51

experience heightened flood risk and significant geomorphological changes. 52

3

Keywords: extreme floods; lichenometry; SNAO; climate change; Arctic amplification 53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

4

1. Introduction 78

79

Large floods in small headwater catchments of the UK are highly effective 80

geomorphological agents and their past occurrence can be readily identified from 81

field evidence, including boulder berms, debris cones, and alluvial fan deposits 82

(Carling, 1986; Harvey, 1986; Wells and Harvey, 1987; Coxon et al., 1989; Macklin 83

et al., 1992; Merrett and Macklin, 1999; Johnson and Warburton, 2002; Macklin and 84

Rumsby, 2007; Milan, 2012; Foulds et al., 2013). Dating of these features is critical 85

to understanding (i) landscape evolution in response to past, present, and future 86

climate change; and (ii) the magnitude–frequency of extreme, ungauged floods in 87

small upland catchments. The latter is especially important as it allows upland flood 88

records to be extended beyond the typical range of instrumental data (ca. 35 years in 89

the UK; Macdonald, 2013). In small, ungauged mountain catchments, combining 90

geomorphological and documentary data offers a reliable way to investigate longer 91

term changes in extreme weather and flood frequency over the past two to three 92

centuries (Maas and Macklin, 2002; Macklin and Rumsby, 2007; Ruiz-Villanueva et 93

al., 2013). Records of this length are important because short instrumental records 94

do not cover flood-rich periods in the nineteenth century and first half of the twentieth 95

century (Bichet et al., 2013; Foulds et al., 2013). This can, in turn, lead to the 96

underestimation of flood risk (Black and Fadipe, 2009). Indeed, following recent 97

large flood events in the UK (e.g., summer 2007, 2012) reports of their 98

‘unprecedented’ nature and ‘biggest in living’ memory are common (Foulds et al., 99

2012). In contrast, analysis of documentary (Macdonald and Black, 2010; 100

Macdonald, 2012, 2013; Pattison and Lane, 2012) and geomorphological records 101

(Macklin and Rumsby, 2007; Foulds et al., 2013) often reveals a different story. That 102

5

is, large floods have occurred frequently in the past associated with flood-rich 103

periods and variability of the NAO, SNAO, and the frequency/persistence of different 104

Lamb weather types (LWTs; Lamb, 1972), notably cyclonic and westerly flows 105

(Rumsby and Macklin, 1994; Longfield and Macklin, 1999; McEwen, 2006; Macklin 106

and Rumsby, 2007; Macdonald, 2012; Foulds et al., 2013; Wilby and Quinn, 2013). 107

108

Although geomorphological methods of flood series extension have been widely 109

used in the British uplands, records for Wales are restricted to a single catchment in 110

the Brecon Beacons (Macklin and Rumsby, 2007) and a series of debris flows in one 111

valley in Snowdonia (Winchester and Chaujar, 2002). The Cambrian Mountains, 112

which form the central uplands of Wales (Fig. 1), have yet to be studied in detail 113

because of their remoteness and difficulties locating and integrating Welsh and 114

English language documentary flood sources. However, the area is susceptible to 115

high rainfall and flooding (Newson, 1975, 1980). Following one such event in June 116

2012 (Foulds et al., 2012), several small headwater valleys in the area appeared to 117

contain significant evidence of geomorphologically effective historical floods (i.e., 118

lichen-covered boulder berms) requiring further investigation. 119

120

The key aims of this study are to (i) elucidate the chronology of these extreme flood 121

events in the Cambrian Mountains; and (ii) identify climatological and meteorological 122

controls on flood generation. The latter aim is central to understanding local to 123

regional scale catchment response to past, present, and future weather extremes, 124

and inherent flood risk. Catchments investigated in this study are much smaller (< 5 125

km2) than typical down-scaled regional climate models (25-50 km2), which have been 126

shown to perform unreliably in small, steep catchments (Smith et al., 2013). These 127

6

are precisely the areas where damaging flash floods occur, and we advocate in this 128

paper a geomorphological approach to better understand future flood risk in 129

catchments of this nature. 130

131

2. Study catchments 132

133

The Cambrian Mountains are located in mid- and west Wales, UK (Fig. 1). They 134

include upland plateaux, typically higher than 300 m Above Ordnance Datum (AOD) 135

(maximum relief is 752 m AOD on Pen Pumlumon Fawr), which are dissected by 136

glacial troughs and the headwaters of some of the largest Welsh rivers, including the 137

Hafren (Severn), Gwy (Wye), Rheidol, Teifi, and Ystwyth. This study focuses on the 138

upper reaches of the Afon Ystwyth (Fig. 1), which rises at 535 m AOD and has a 139

catchment area of 191 km2. In the lowlands, at Pont Llolwyn, Q95 and Q10 are 0.60 140

and 14.33 m3 s-1 (drainage area = 170 km2), respectively, compared to 0.18 and 5.60 141

m3 s-1 in the uplands at Cwmystwyth (drainage area = 32 km2) (Afon Ystwyth gauged 142

data are available through National River Flow Archive: www.ceh.ac.uk). Flood 143

hydrographs in small catchments of the Cambrian Mountains are typically very 144

flashy, with short lag times (< 2.5 hours; Newson, 1975). 145

146

The climate of the Ystwyth catchment is typically mild and wet. Mean annual rainfall 147

is 1876 mm at Cwmystwyth (301 m AOD) and 1217 mm at Trawsgoed (63 m AOD). 148

Mean annual temperature is 9.9°C in the lowlands, and average annual 149

temperatures at Cwmystwyth are 7.1–9.2°C (meteorological data for the Ystwyth 150

catchment are available through the Met Office: www.metoffice.gov.uk). The 151

Cambrian Mountains tend to be affected by frontal rainfall associated with Atlantic 152

7

depressions at any time of the year, as well as slow moving, convective summer 153

storms (Newson, 1975, 1980). An average winter may also see several short-lived 154

snow accumulation and subsequent melt events and ca. 60 days of air frost per year 155

at Cwmystwyth. Geologically, the area is dominated by Silurian deposits of the 156

Llandovery series (shales, siltstones, sandstones, and mudstones of the 157

Cwmystwyth Grits Formation). 158

159

This study concentrates on a series of boulder berms (Figs. 2A-2C), alluvial fans, 160

and debris cone deposits (Fig. 2D) that were identified within the small, steep, 161

headwater tributaries of Nant Cwm-Du (Lat. 52o21’18’’ N., Long. 3o44’40’’ W.; area = 162

0.9 km2, average channel bed slope = 0.132 m m-1), Nant Gau (Lat. 52o19’57’’ N., 163

Long. 3o47’37’’ W.; area = 2.7 km2, average channel bed slope = 0.079 m m-1), and 164

Nant Milwyn (Lat. 52o20’18’’ N., Long. 3o46’05’’ W.; area = 3.7 km2, average channel 165

bed slope = 0.096 m m-1), which are northward-draining, left (south) bank tributaries 166

of the Afon Ystwyth, as well as several berm and cone deposits in the upper reaches 167

of the Rheidol and Leri catchments. 168

169

3. Methods 170

3.1. Lichenometry 171

172

Lichenometry is a standard and widely used dating technique in geomorphology. 173

Although originally developed to date glacial deposits (Beschel, 1961), the technique 174

has been applied with considerable success to constraining the age of boulder-berm 175

flood deposits in the British uplands (Macklin et al., 1992; Macklin and Rumsby, 176

2007). In the Cambrian Mountains lichenometry was used to date boulder berms 177

8

that were identified using a combination of aerial photographs, ground truthing, and 178

geomorphological field mapping at a scale of 1:10, 000. Once identified, all boulders 179

per berm were searched for lichens. A combination of Porpidia tuberculosa and 180

Rhizocarpon geographicum was used to date surfaces because the former grows 181

relatively quickly (typically > 0.5 mm y-1) and gives high resolution dating for deposits 182

< 75 years in age; beyond this, because of the relatively fast growth rate, lichens 183

tend to coalesce, which prevents measurement. For older deposits, the slower 184

growing R. geographicum (typically < 0.5 mm y-1) is capable of giving better dating 185

control, although coalescing lichens and general vegetation growth can still be 186

problematic. In this study, these problems were most apparent in the Nat Gau 187

catchment, where a series of terraced berms (4-5 m above the present channel) 188

could not be dated. 189

190

An indirect dating method was used, which relates the size of lichen growing on 191

surfaces of unknown age (e.g., flood deposits) to empirical lichen size–age 192

relationships derived for the same lichen on surfaces of known age (e.g., 193

gravestones) (Macklin et al., 1992; Merrett and Macklin, 1999; Foulds et al., 2013). 194

Local graveyards at Hafod (2-5 km from the study sites) and Ysbyty Ystwyth (5-9 km 195

from the study sites; Fig. 1) were used to construct lichen size–age relationships. 196

The most commonly used lichenometry methods, especially in fluvial studies 197

(Macklin and Rumsby, 2007), involve measurement of the single largest lichen (LL) 198

or the mean of the three or five largest specimens (3LL, 5LL), respectively. For 199

details of other lichenometry methods available to practitioners (e.g., fixed area 200

largest lichen, size frequency, and percentage coverage) see Bradwell (2009). 201

202

9

In order to test which method is the most reliable, calibrated size–age relationships 203

were constructed for LL, 3LL, and 5LL methods. The validity of these lichen size–204

age relationships was tested by collecting a second batch of P. tuberculosa lichen 205

data from independent graveyards and predicting tombstone ages based on the 206

linear regression equations shown in Fig 3A. These dates could then be compared 207

to the true (inscribed) ages (Fig. 3B; Table 1). The 3LL method was selected 208

because of the slightly higher percentage of predictions accurate to > 10 years 209

compared to the 5LL method and the greater accuracy dating older surfaces 210

compared to the LL method (Table 1). The Rhizocarpon size–age relationship could 211

not be validated because gravestones with at least three lichens growing on them 212

could not be found in the test graveyards; this was probably owing to differences in 213

microclimate between graveyard sites. In this, study all lichen ages are plotted 214

showing ± 2 σ error bars based on test data for P. tuberculosa (Table 1). 215

216

3.1.1 Lichen growth rates 217

218

Average, minimum, and maximum growth rates for P. tuberculosa in the Cambrian 219

Mountains are 1.20, 0.73, and 2.28 mm y-1 and 0.42, 0.34, and 0.67 mm y-1 for R. 220

geographicum. These rates were calculated by dividing average lichen size by 221

gravestone age. The latter was adjusted for lag colonisation periods of ca.10 (P. 222

tuberculosa) and ca.15 (R. geographicum) years, respectively. These rates are 223

typical for P. tuberculosa and R. geographicum growing in upland Britain (see 224

Armstrong and Bradwell, 2010; Foulds et al., 2013). Images of these lichen species 225

and habitat details can be found on the British Lichens website 226

(www.britishlichens.co.uk) and in Dobson (2011). 227

10

3.2. Palaeohydrological analysis 228

229

Relative flood magnitude was assessed by measuring the mean B-axis of the five 230

largest boulders (Di) present in each boulder-berm unit. A variety of boulder 231

transport equations can be used to estimate palaeohydrological characteristics, the 232

most commonly used being those of Costa (1983), based on data from the Colorado 233

Front Range. Using this method, average velocity (V; Eq. 1) is calculated from 234

average boulder dimensions (Di), and this figure is then multiplied by cross-sectional 235

area to estimate discharge. Perhaps more appropriate in a UK context is the method 236

of Carling (1986), developed from field data in a steep upland catchment in northern 237

England (Eq. 2; Carling, 1983, 1986), an environment similar to mid-Wales, although 238

the streams investigated in the Cambrian Mountains are steeper than those studied 239

by Carling (1983, 1986). In Eq. (2), A, γ, S and n represent cross-sectional wetted 240

area (estimation of palaeostage is described below), specific weight of water, 241

average channel bed slope, and Manning’s roughness coefficient, respectively. 242

243

V = 0.18 Di 0.487 (1) 244

Q = 3.06 x 10-2 A Di -2/9 γ-2/3 S-1/6 n-1 (2) 245

n = 0.32 S0.38 R-0.16 (3) 246

247

Palaeostage was estimated using berm crest elevation (Carling, 1987; Kehew et al., 248

2010) and the area between the berm base/floodplain contact and berm crest was 249

estimated as the depth of overbank flow (used to approximate cross-sectional area 250

of flow). Variables of S and A were based on field survey using a differential GPS 251

(Trimble R8), and n values used were 0.05 and 0.07 (i.e., Chow’s (1959) normal and 252

11

maximum values for boulder streams). Roughness was also derived empirically 253

from S and R (hydraulic radius) (Eq. 3; Jarrett, 1992). The latter method typically 254

gives much higher n values, in this case 0.13-0.15, similar to other published values 255

for UK mountain streams (Johnson and Warburton, 2002). Palaeodischarges were 256

only estimated for the Nant Cwm-Du, which has a very simple valley floor relief and 257

morphology (i.e., berms deposited on straight floodplains) suitable for discharge 258

reconstructions. The majority of berms in the other study catchments are located on 259

terraces and associated with palaeochannels, not the modern channel; and thus too 260

many assumptions would need to be made about pre-flood cross-sectional areas to 261

arrive at reliable estimates of discharge. 262

263

3.3. Climatological and documentary data 264

265

Some of the largest known floods in the British uplands have been caused by near 266

stationary summer thunderstorms that can deliver 100 to > 200 mm of rain in 2-5 267

hours (Burt, 2005). Atmospheric conditions most favourable to these situations are 268

often associated with meridional flow patterns when the Summer North Atlantic 269

Oscillation (SNAO) (Folland et al., 2009) in its negative mode, which pushes the jet 270

stream and the associated Atlantic storm track farther south than usual (Dong et al., 271

2013). Folland et al.’s (2009) ‘high summer’ (July-August) SNAO index data were 272

used to explore longer term changes in summer atmospheric circulation and its 273

potential impact on flood frequency. For the winter half year (October-March), 274

station-based NAO index data were used to examine climate-flood linkages 275

(downloaded from the University of East Anglia’s Climatic Research Unit: 276

12

www.cru.uea.ac.uk). Known flood dates were also analysed with respect to LWTs 277

based on Jones et al. (2012b). 278

279

A documentary flood history (Table 2) was compiled from English and Welsh 280

language newspaper archives and local school logbooks (Parry, 2013). The value of 281

such sources, especially ‘grey literature’ (Uhlemann et al., 2013), for reconstructing 282

historical climatology and geomorphology has been shown by Brazdil et al. (2006), 283

McEwen and Werritty (2007), and Macdonald et al. (2010). The latter study 284

highlighted the as yet untapped potential of historical documents written in Welsh, 285

particularly where they can be integrated with English-medium sources and 286

instrumental series. However, documentary flood references are often subjective, 287

can exaggerate magnitude, and/or mis-date events (Foulds et al., 2012, 2013). 288

Furthermore, flood impacts are often reported only for large rivers and specifically for 289

infrastructure/property flooding. This can be problematic because a flood that has 290

been documented need not necessarily have been geomorphologically effective. In 291

remote mountain catchments, floods may also go unreported. These factors mean 292

that caution should be taken to not overinterpret historical sources. 293

294

4. Results 295

4.1. Geomorphological-based flood chronology 296

297

Figure 4 shows decadal summaries of the total number of channel and slope floods 298

in the Nant Cwm-Du (Fig. 4A), Nant Gau (Fig. 4B) and Nant Milwyn (Fig. 4C) 299

catchments. Overall, these data indicate that the most geomorphologically effective 300

known hydrometeorological conditions occurred in the Cambrian Mountains during 301

13

the period 1900-1960 (Fig. 4D), although there are variations between the three 302

study catchments. Peak geomorphological activity occurred in the Nant Cwm-Du 303

system in the 1920s (7 berms) and from 1940 to 1960 (14 berms). These dates are 304

similar to the Milwyn catchment, where peak berm deposition occurred in the 1910s 305

(6 units) and two berms in a small tributary of the upper Rheidol (Nant y Llyn) (1905 306

±10 and 1945 ±10, respectively). A single berm was also dated to the 1932 ±10 in 307

the upper Leri catchment. In contrast, berms were deposited slightly earlier in the 308

Nant Gau catchment (1860s to 1880s: 2-3 units; 1920s: 3 units). Older deposits in 309

the Nant Gau system correspond with a wider valley floor and a series of terraced 310

berms (Fig. 2C) that predate more recent (floodplain) deposits. Our data suggest 311

that preservation potential in the narrow Nant Cwm-Du and Nant Milwyn catchments 312

is lower than in the Nant Gau catchment because of high rates of valley floor 313

reworking. This means that the alluvial record has been censored (Lewin and 314

Macklin, 2003) and skewed toward the twentieth century; Foulds et al. (2013) 315

reported a similar finding in catchments of < 1.5 km2 on Dartmoor. Low ‘apparent’ 316

rates of deposition before 1900 are also a reflection of berms that could not be dated 317

owing to coalescing lichens on older deposits. 318

319

In terms of slope activity (Figs. 2D, 4D), the majority of dates correspond with berm 320

deposition between 1900 and 1940, as well as a shallow gully and associated debris 321

cone cut into the back wall of the Cwm dated to 1991 ±10. A gully and debris fan 322

generated by an event in June 2012 was also identified on the Nant yr Hulog, which 323

is a very steep (0.27 m m-1) tributary in the upper Rheidol catchment. Most of the 324

slope failures in June 2012 were disconnected from river channels (see Foulds et al., 325

2012); this partly explains the small number of berms generated during this event. 326

14

Field evidence strongly suggests that berm generation in the study catchments is 327

linked to strong slope-channel coupling through undercutting and bank/slope 328

collapse. In all of the catchments, but especially in the Nant Cwm-Du and Nant Gau, 329

a series of large, vegetated erosional scars indicate past episodes of erosion; some 330

of the most extensive berm deposits can be found downstream of these features 331

(Fig. 5). Low rates of berm deposition in recent decades (post-1960) are mirrored by 332

a lack of recent evidence of high sediment supply from boulder-rich drift deposits. 333

Lichen data suggest that the last large-scale valley floor reactivation phase probably 334

took place sometime in the 1940s/1950s. 335

336

4.2. Linking documentary and geomorphological flood records 337

338

Our lichen-based chronology points to frequent berm deposition between 1900 and 339

1960. However, because of dating accuracy limitations, many berm ages overlap, 340

making it difficult to be precise about the timing of flood events. To refine the lichen 341

chronology, documentary sources have been used in an attempt to identify specific 342

flood events (Table 2). Many of the largest floods in small upland UK catchments 343

have been associated with slow moving summer storms (e.g., West Yorkshire, 1944: 344

Doe and Brown, 2005; Lynmouth, 1952: Dobie and Wolf, 1953; North Pennines, 345

1983: Carling, 1986; Howgill Fells, 1982: Harvey, 1986; Boscastle, 2004: Roca and 346

Davison, 2009). Based on these known flood-triggering conditions, documented 347

events in June 1910, July 1926, June 1931, June 1935, and July/August 1957 (Table 348

2), all of which refer to torrential thundery rain, are the most likely to have been 349

responsible for boulder-berm deposition in the early to mid-twentieth century. Some 350

especially ‘flood-rich’ years were 1886, 1909, 1910, and 1957 (Table 2); all of these 351

15

years correspond with decades identified as having high rates of berm deposition 352

(Fig. 4D). Additionally, taking dating errors into account, repeated flooding in 1957 353

(Table 2) may explain the valley floor reactivation phase sometime in the 354

1940s/1950s. In terms of slope activity, the deepest gullies and associated debris 355

fans in the Ystwyth headwaters (e.g., Fig 2D) also date to between 1905 ±10 and 356

1931 ±10; documentary sources also record a large landslide in the upper Ystwyth 357

valley in September 1922 (Table 2). These dates suggest the potential for strong 358

slope-channel coupling in the early twentieth century. 359

360

A boulder-berm dated to 1932 ±10 in the upper Leri catchment also corresponds with 361

a severe thunderstorm and flood in June 1935 (Table 2). The potential 1935 berm is 362

significant because at the same site and inset on a lower alluvial unit (floodplain), 363

there is a boulder-berm deposit from June 2012, which has almost identical mean 364

and maximum boulder dimensions to 1935 material (Fig. 6). These data confirm that 365

the June 2012 flood was not ‘unprecedented’, having very likely occurred at least 366

twice in ca. 77 years. 367

368

4.2.1 Flood seasonality 369

370

Detailed seasonal analysis of documentary events in the Cambrian Mountains (Fig. 371

7) shows a sharp increase in flood frequency between 1870 and 1925 (autumn) and 372

from 1870 to 1930 (summer). Winter floods were most common between 1900 and 373

1910. After 1910 and 1925, autumn and winter flooding declined, although summer 374

flooding continued intermittently up to the 1970s. In comparing berm and 375

documentary flood series, summer and autumn events show the best agreement 376

16

with high rates of berm sedimentation in the late nineteenth and early to mid-377

twentieth centuries. 378

379

4.3. Palaeohydrology of flood events 380

4.3.1. Relative flood magnitude 381

382

Figure 8 shows proxy flow magnitude based on average boulder dimensions. These 383

data show a marked reduction in flood magnitude between 1850 and present. The 384

largest boulders (0.7-0.9 m) are associated with terraced deposits in the Nant Gau 385

catchment dating to the mid-to-late nineteenth century. Two floods of comparable 386

size to nineteenth century events took place in 1923 ±10 and 1949 ±10. After ca. 387

1960, flood magnitude declined (average boulder dimensions < 0.5 m; Fig. 8). An 388

important caveat is that using boulder data in this way assumes an unlimited supply 389

of boulders of all sizes through time. If earlier floods had exhausted the largest 390

boulders, there would be a limit to the maximum calibre of sediment available for 391

transport during later floods. This scenario would give an impression of decreasing 392

flood magnitude. However, this is not thought to be the case in the study 393

catchments because (i) large boulders are abundant in all of the present day river 394

channel beds; (2) in the small, narrow catchments of the Cambrian Mountains, large 395

floods would be easily capable of stripping and remobilising boulders associated with 396

historical berm units (e.g., Milan, 2012); and (iii) all of the rivers investigated have a 397

high potential for strong slope-channel coupling as they are deeply incised through 398

boulder-rich drift deposits. 399

400

401

17

4.3.2. Discharge 402

403

Table 3 shows discharge estimates for 29 flood berms in the Nant Cwm-Du 404

catchment based on Carling (1986), Costa (1983), and a variety of n values. Notable 405

differences between these Q estimates mean that it is necessary to assess which, if 406

any, are the most realistic. Table 4 shows previously published discharge estimates 407

from other small UK catchments, and Fig. 9 shows data from Nant Cwm-Du plotted 408

alongside other UK and European floods. For systems of < 2 km2 in an upstream 409

catchment area, Table 4 suggests that peak discharges of 4-7 m3 s-1 are realistic. 410

Combining Carling’s (1986) method of estimating discharge (Eq. 2) with n values 411

derived empirically using Jarrett (1992; Eq. 3) produces similar peak flow estimates 412

(1-7 m3 s-1; Table 3). Table 4 and Fig. 9 also suggest that for the slightly larger Gau 413

and Milwyn catchments, formative discharges may have been in the region of 20-40 414

m3 s-1. On the upper Leri, Foulds et al. (2012) estimated the 2012 flood at between 415

21 and 31 m3 s-1 (area = 7 km2). 416

417

Costa’s (1983) method produced much higher Q estimates (Table 3), as reported in 418

previous research (Carling, 1986; Johnson and Warburton, 2002). These 419

overestimates occur because average water velocities calculated using Costa (1983) 420

are relatively high (> 4 m s-1) and exceed typical flash flood velocities of < 3 m s-1 421

(Marchi et al., 2010; Lumbroso and Gaume, 2012). Carling (1986) estimated an 422

average water velocity of 1.2 m s-1 for the Noon Hill event based on the observed 423

flood wave travel time. This compared to velocities > 4 m s-1 based on Costa (1983). 424

425

426

18

4.4. Flood generation and synoptic conditions in the Cambrian Mountains 427

428

Table 5 shows the meteorology of documentary floods in the Cambrian Mountains 429

based on LWTs and SNAO/NAO data for three days before and on the flood day. 430

The majority (52%) of historical flood events were associated with cyclonic flow (or 431

some variant thereof; e.g., cyclonic westerly). Other weather types associated with 432

flooding include southerly (19%), westerly (10%), northerly, easterly, and anticyclonic 433

variants (all 6% each) (Table 5). For summer (July-August) events, the strongest 434

control on rainfall and discharge at Cwmystwyth is the SNAO (Fig. 10). Negative 435

SNAO index values correlate with wetter, cooler, and cloudier than average 436

conditions and vice versa for positive index values (Folland et al., 2009). Indeed, 437

many documented summer floods were associated with negative SNAO index 438

values (Table 5). An important exception occurs when warm anticyclonic conditions 439

and positive SNAO index values (associated with the Azores high over the UK) 440

breakdown and very humid air is drawn in from the near continent, sparking severe 441

thunderstorms (e.g., August 1957, 1977: Table 5; June 1910: Table 2). 442

443

Monthly NAO values indicate that historical autumn-winter floods occurred during 444

negative phases associated with cyclonic conditions (Table 5), which can give rise to 445

circulating frontal rain bands and documentary references to rainfall continuing over 446

several days (e.g., February 1869, December 1880, and September 1903: Table 2). 447

Positive autumn-winter NAO values are also associated with an increased frequency 448

of westerly/southwesterly winds, which can lead to orographically enhanced daily 449

and multiday rainfall totals in upland areas (Burt, 2005; Burt and Howden, 2013) 450

(e.g., November 1894, 1929, December 1979, March 1998, October 2000: Table 2). 451

19

Positive NAO conditions correlate well with average autumn-winter flows at 452

Cwmystwyth and maximum flows in December (Table 6). However, these data 453

should be treated with caution because the gauge record at Cwmystwyth is very 454

short and covers a period of frequent, low magnitude events. Indeed, Table 5 455

suggests that many historical autumn-winter floods were associated with negative 456

NAO conditions. 457

458

4.5. Longer term climatic context of extreme floods in the Cambrian Mountains 459

460

In the light of strong correlations between the SNAO, cyclonic weather types, rainfall, 461

river flows, and known extreme flood-generating mechanisms (i.e., summer storms), 462

the full SNAO and cyclonic LWT series can be used as useful proxies for past 463

periods of high rainfall and flooding (Fig. 11), including summer 2012. Figures 11A 464

and 11B show that from 1850 to 1900, from 1910 to 1940, and during the 1950s 465

there were notable negative SNAO anomalies, high cyclonic activity, and rainfall in 466

SW England and Wales. These climatic data correspond well with summary 467

boulder-berm data in terms of peak deposition in the early to mid-twentieth century 468

and peak magnitude in the second half of the nineteenth century and in the ca. 469

1920s and ca. 1940s. The full SNAO, cyclonic LWT, and SW England and Wales 470

rainfall series provide strong explanations for the sharp decline in boulder-berm 471

deposition and flood magnitude after ca. 1960. After this date, the frequency of 472

positive SNAO anomalies increased, accompanied by a sharp decline in the 473

frequency of cyclonic summer flow and rainfall. Additionally, Fig. 11A highlights the 474

negative shift in the SNAO since 2007, which corresponds with a run of very wet 475

20

summers in the UK and widespread flooding (including June 2012 in the Cambrian 476

Mountains). 477

478

During the autumn and winter, there is some correspondence between the timing of 479

positive NAO phases, high rainfall, documented floods, and berm deposition, 480

especially from 1910 to 1960 (Fig. 11C). However, the average Nov-Feb NAO does 481

not provide an adequate explanation for the sharp decline in boulder-berm 482

sedimentation and flood magnitude in Wales and the wider British uplands during the 483

late twentieth century (Fig. 12; Macklin and Rumsby, 2007). On the contrary, 484

declining berm sedimentation after ca. 1960 in small mountain catchments 485

corresponds to above average autumn-winter rainfall and a well-documented run of 486

positive NAO index values (Fig. 11C; Hurrell and van Loon, 1997; Osborn, 2006). 487

488

5. Discussion 489

5.1. Comparisons with other UK lichen-based flood records 490

491

The timing of geomorphologically effective flood events in the Cambrian Mountains 492

has many similarities with previously published lichen-based flood chronologies in 493

the British uplands (Fig. 12). Most notably, geomorphic activity in the Brecon 494

Beacons (south Wales) shows a similar early twentieth century peak, and 495

Winchester and Chaujar (2002) reported peak debris flow activity in Snowdonia 496

(north Wales) between the 1880s and late 1920s, similar to mid-Wales. There is 497

also very good agreement between high rates of berm deposition on Dartmoor 498

(Foulds et al., 2013), in the North Pennines (Macklin et al., 1992), and Yorkshire 499

Dales (Merrett and Macklin, 1999) in the late nineteenth and early twentieth 500

21

centuries. In terms of flood magnitude, average boulder dimensions were generally 501

declining throughout the 1900s in mid-Wales, very similar to downward trends in 502

flood competence during the twentieth century in other small upland catchments 503

(Macklin et al., 1992; Merrett and Macklin, 1999; Foulds et al., 2013). The abrupt 504

late twentieth century decline in flood activity in the British uplands is probably the 505

greatest area of commonality in all of the respective flood records (Fig. 12). 506

507

5.2. Climatological controls on upland flooding 508

5.2.1. Summer 509

510

The timing of geomorphologically effective floods in the British uplands and their 511

large-scale atmospheric context appear to be controlled by negative phases of the 512

SNAO and associated southerly migration of the North Atlantic storm track toward 513

the UK (Folland et al., 2009; Dong et al., 2013). Under these conditions, meridional 514

flow patterns, cyclonic circulation, extreme rainfall, and floods are more frequent 515

(Rumsby, 1991; Macklin et al., 1992; Rumsby and Macklin, 1994; Foulds et al., 516

2013). Extreme rainfall leads to high rates of surface runoff, which activate hillslope 517

sediment stores and increase sediment supply, owing to effective slope-channel 518

coupling in small, narrow catchments with limited valley floor storage space (Harvey, 519

1986; Wells and Harvey, 1987). 520

521

In contrast, periods of reduced upland geomorphic activity during the late twentieth 522

century (Macklin and Rumsby, 2007) are related to a transition to above average 523

SNAO anomalies since the ca. 1960s, indicating a change toward persistent 524

anticyclonic flow during recent decades (up to ca. 2007) (Linderholm et al., 2008). 525

22

Indeed, many instrumental records show reduced summer rainfall and intensity from 526

1961 to 2000 (Osborn and Hulme, 2002; Wilby et al., 2008; Biggs and Atkinson, 527

2011; Burt and Ferranti, 2012; Jones et al., 2012a). These dry summer conditions 528

are clearly manifest in the British boulder-berm flood record (Fig. 12), suggesting that 529

although winter floods, especially when combined with snowmelt (Rumsby and 530

Macklin, 1994, 1996; Merrett and Macklin, 1999; Johnson and Warburton, 2002), can 531

initiate appreciable geomorphic changes, the UK boulder-berm record would appear 532

to be a useful proxy of summer climate (high rates of berm deposition = wetter than 533

average summers; low rates of berm deposition = drier than average summers). 534

535

5.2.2. Autumn-winter 536

537

Winter NAO (Nov-Feb) variability appears to have had less of an impact on small UK 538

mountain catchments because (i) long duration rainfall events generally have lower 539

intensities than short duration convective storms, and (ii) return periods only tend to 540

become extreme for periods > 24 hours (Archer et al., 2005; Walsh, 2010). Whilst 541

capable of wetting slopes to the point of failure, this type of rainfall cannot produce 542

the extreme flood peaks that transform river channels in small catchments (Newson, 543

1989). In contrast, autumn-winter frontal rainfall during the 1980s and 1990s caused 544

a notable flood-rich period in large UK catchments (McEwen, 2006; Macdonald, 545

2012; Pattison and Lane, 2012), owing to a prolonged run of positive NAO values 546

(Hurrell, 1995; Hurrell and van Loon, 1997). Asynchrony in the timing of extreme 547

floods in the uplands and lowlands is important because it implies that flood risk 548

associated with future climatic change will vary spatially and temporally, even with 549

different parts of the same river basin. 550

23

5.3. Climate change and future floods in the Cambrian Mountains 551

552

Smith et al. (2013) suggested that the much quoted ‘wetter autumn-winters/drier 553

summers’ climate change scenario for the UK may be less clear-cut than first 554

thought, and they highlight appreciable variations from catchment-to-catchment, 555

season-to-season, and between seasonal averages and extremes. Most notably, 556

the ‘drier/warmer’ scenario is being complicated by the possible role of Arctic 557

amplification and rapid sea ice loss, which may be directly affecting mid-latitude 558

weather systems (Francis and Vavrus, 2012; Overland et al., 2012; Screen, 2013; 559

Screen and Simmonds, 2013). Specifically, Arctic amplification reduces the equator-560

pole temperature gradient, which leads to enhanced meridional flow (Overland et al., 561

2012; Screen, 2013) and causes a southward displacement of the jet stream toward 562

the UK. These conditions favour unusually wet summers in the UK (e.g., 2007-2012) 563

and boulder-berm generation. 564

565

Evidence of the ‘wetter autumn/winters’ scenario is apparent in many short 566

instrumental records (Wilby et al., 2008). Although Biggs and Atkinson (2011) 567

reported increasing hydrological extremes in parts of the Cambrian Mountains, the 568

term ‘extreme’ is relative because the period covered in their study (1977-2006) does 569

not include any large historical events (e.g., Table 2). Flow records beginning in the 570

relatively dry 1970s are ‘hard-wired’ to show increased flood risk (Wilby and Quinn, 571

2013); if longer records were available, the results of Biggs and Atkinson (2011) 572

might turn out to be less significant. In the absence of long instrumental records, 573

other evidence must be used to extend climate/flood records. Geomorphological 574

data, combined with documentary data and longer term atmospheric proxies, are key 575

24

components to better understanding present and future flood risk in small upland 576

catchments. Indeed, ungauged mountain catchments are a global phenomenon and 577

we are confident that the methods outlined in this paper could be applied beyond the 578

UK or Europe. 579

580

6. Conclusions 581

582

In upland areas of the UK small, steep, boulder-bed streams are often ungauged. 583

This poses a particular problem because these types of catchment are capable of 584

generating severe floods, with serious consequences for local infrastructure and 585

public safety (e.g., bridge collapse, dam wall integrity). Investigations in mid-Wales 586

showed that coarse-grained flood deposits can be dated using lichenometry to 587

provide a valuable method of extending upland flood records (> 150 years in length). 588

Data sets of this length are critical to fitting short-term instrumental series into their 589

longer term context. Our new data from mid-Wales, combined with previously 590

published data, show that the incidence of extreme flooding in small upland 591

catchments is best explained by the SNAO and shifts between negative (wetter than 592

average conditions with regular cyclonic flow and flooding) and positive phases (drier 593

than average conditions with less frequent cyclonic flow and flooding), which vary 594

from individual summers to decadal and multidecadal periods. Recent wet summer 595

weather, flooding, and boulder-berm deposition in the UK (2007-2012) are related to 596

a pronounced negative phase shift of the SNAO. 597

598

Evidence suggests that recent summer weather extremes (2007–2012) in the mid-599

latitudes are related to Arctic amplification and rapid sea-ice loss, which favours 600

25

enhanced meridional flow patterns and flooding. This means that future climatic 601

change and accelerated Arctic melting may lead to (i) heightened flood risk, and (ii) 602

appreciable geomorphic changes. In turn, this may also lead to potential sediment 603

management issues in headwater catchments. Finally, based on recent evidence 604

that down-scaled regional climate models perform unreliably in small, mountainous 605

systems (Smith et al., 2013), geomorphologists are well placed to advance our 606

understanding of future climatic changes and their potential impacts on flood risk and 607

landscape change. 608

609

Acknowledgements 610

611

We are grateful to Rhodri Bevan for his valuable field assistance and to the staff at 612

Ceredigion Archives (Aberystwyth), as well as Sioned Llywelyn, who collected 613

several School Log Book records. Digital copies of newspapers were accessed 614

through the National Library of Wales’ Welsh Newspapers Online service available 615

at http://papuraunewyddcymru.llgc.org.uk/en/home. Professor Chris Folland (Met 616

Office) provided the SNAO data. We would also like to thank Dr Stephen Tooth and 617

the reviewers, whose helpful comments improved the manuscript. 618

619

620

621

622

623

624

625

626

26

References 627

628

Acreman, M.C., 1989. Extreme historical UK floods and maximum flood estimation. Journal 629

of the Institute of Water and Environmental Management 3, 404-412. 630

631

Alexander, L.V., Jones, P.D., 2001. Updated precipitation series for the UK and discussion 632

of recent extremes. Atmospheric Science Letters 1, 142-150. 633

634

Archer, D.R., Leesch, F., Harwood, K., 2005. Learning from the extreme River Tyne flood in 635

January 2005. Water and Environment Journal 21, 133-141. 636

637

Armstrong, R., Bradwell, T., 2010. Growth of crustose lichens: a review. Geografiska 638

Annaler: Series A, Physical Geography 92, 3-17. 639

640

Beschel, R.E., 1961. Dating rock surfaces by lichen growth and its application to glaciology 641

and physiography (lichenometry). In: Raasch, G.O. (Ed.), Geology of the Arctic: Proceedings 642

of the First International Symposium on Arctic Geology, Vol. 2. University of Toronto Press, 643

Toronto, Canada, pp. 1044-1062. 644

645

Bichet, A., Folini, D., Wild, M., Schar, C., 2013. Enhanced central European summer 646

precipitation in the late nineteenth century, a link to the Tropics. Quarterly Journal of the 647

Royal Meteorological Society, DOI:10.1002/qj.2111. 648

649

Biggs, E.M., Atkinson, P.M., 2011. A characterisation of climate variability and trends in 650

hydrological extremes in the Severn Uplands. International Journal of Climatology 31, 1634-651

1652. 652

653

27

Black, A.R., Fadipe, D., 2009. Use of historic water level records for re‐assessing flood 654

frequency, case study of the Spey catchment. Water and Environment Journal 23, 23-31. 655

656

Bradwell, T., 2009. Lichenometric dating: a commentary, in the light of some recent 657

statistical studies. Geografiska Annaler: Series A, Physical Geography 91, 61-69. 658

659

Brazdil, R., Kundzewicz, Z.W., Benito, G., 2006. Historical hydrology for studying flood risk in 660

Europe. Hydrological Sciences Journal 51, 739-764. 661

662

Burt, S., 2005. Cloudburst upon Hendraburnick Down, The Boscastle Storm of 16th August 663

2004. Weather 60, 219-227. 664

665

Burt, T., Ferranti, E.J.S., 2012. Changing patterns of heavy rainfall in upland areas, a case 666

study from northern England. International Journal of Climatology 32, 518-532. 667

668

Burt, T.P., Howden, N.J.K., 2013. North Atlantic Oscillation amplifies orographic precipitation 669

and river flow in upland Britain. Water Resources Research 39, 3504-3515. 670

671

Carling, P.A., 1983. Threshold of coarse sediment transport in broad and narrow natural 672

streams. Earth Surface Processes and Landforms 8, 1-18. 673

674

Carling, P.A., 1986. The Noon Hill flash floods, July 17th 1983. Hydrological and 675

geomorphological aspects of a major formative event in an upland landscape. Transactions 676

of the Institute of British Geographers 11, 105-118. 677

678

Carling, P.A., 1987. Hydrodynamic interpretation of a boulder-berm and associated debris-679

torrent deposits. Geomorphology 1, 53-67. 680

681

28

Chow, V.T., 1959. Open Channel Hydraulics. McGraw Hill, New York. 682

683

Costa, J.E., 1983. Palaeohydraulic reconstruction of flash-flood peaks from boulder deposits 684

in the Colorado Front Range. Geological Society of America Bulletin 94, 986-1004. 685

686

Coxon, P., Coxon, C.E., Thorn, R.H., 1989. The Yellow River (County Leitrim, Ireland) flash 687

flood of June 1986. In: Beven, K., Carling, P. (Eds.), Floods, Hydrological, Sedimentological 688

and Geomorphological Implications. Wiley, Chichester, UK, pp. 199-217. 689

690

Dobie, C.H., Wolf, P.O., 1953. The Lynmouth flood of August 1952. Proceedings of the 691

Institute of Civil Engineers 2, 522-588. 692

693

Dobson, F.S., 2011. Lichens: An Illustrated Guide to the British and Irish Species. 6th ed. 694

Richmond Publishing, Slough, UK. 695

696

Doe, R.K., Brown, P.R., 2005. A sea on the moors: devastating flash flooding at Holmfirth, 697

West Yorkshire, UK, May 1944. Journal of Meteorology 30, 163-173. 698

699

Dong, B., Sutton, R.W., Woolings, T., Hodges, K., 2013. Variability of the North Atlantic 700

summer storm track: mechanisms and impacts on European climate. Environmental 701

Research Letters 8 (034037). DOI: 10.1088/1748-9326/8/3/034037. 702

703

Environment Agency, 2013. High flows UK: Wye at Cefn Brwyn (http://www.environment-704

agency.gov.uk/hiflows/station.aspx?55008). Accessed 23/01/14. 705

706

Folland, C.K., Knight, J., Linderholm, H.W., Fereday, D., Ineson, S., Hurrell, J., 2009. The 707

summer North Atlantic Oscillation, past, present, and future. Journal of Climate 22, 1082-708

1103. 709

29

Foulds, S.A., Brewer, P.A., Macklin, M.G., Betson, R., Rassner, S.M.E., 2012. Causes and 710

Consequences of a Large Summer Storm and Flood in west Wales, 8th-9th June 2012. Fluvio 711

Consultancy Report 2012/01/73, Aberystwyth, 53 pp. 712

713

Foulds, S.A., Macklin, M.G., Brewer, P.A., 2013. The chronology and hydrometeorology of 714

catastrophic floods on Dartmoor, southwest England. Hydrological Processes 28 (7), 3067-715

3087. 716

717

Francis, J.A., Vavrus, S.J., 2012. Evidence linking Arctic amplification to extreme weather in 718

mid‐latitudes. Geophysical Research Letters 39 (6), L06801. DOI:10.1029/2012GL051000. 719

720

Harvey, A.M., 1986. Geomorphic effects of a 100 year storm in the Howgill Fells, northwest 721

England. Zeitschrift für Geomorphologie 30, 71-91. 722

723

Hurrell, J.W., 1995. Decadal trends in the North Atlantic Oscillation, regional temperatures 724

and precipitation. Science 269, 676-679. 725

726

Hurrell, J.W., van Loon, H., 1997. Decadal variations in climate associated with the North 727

Atlantic oscillation. Climatic Change 36, 301-326. 728

729

HYDRATE, 2008. HYDRATE Hydrometeorological Data Resources and Technologies for 730

Effective Flash Flood Forecasting (http://www.hydrate.tesaf.unipd.it/). Accessed 10/09/13. 731

732

Jarrett, R.D., 1992. Hydraulics of mountain rivers. In: Yen, B.C. (Ed.), Channel Flow 733

Resistance, Centennial of Manning’s Formula. Water Resource Publications, Littleton, 734

Colorado, pp. 287-298. 735

736

30

Johnson, R.M., Warburton, J., 2002. Flooding and geomorphic impacts in a mountain 737

torrent, Raise Beck, central Lake District, England. Earth Surface Processes and Landforms 738

27, 945-969. 739

740

Jones, C.A., 2013. Human-environment interactions during periods of extreme weather in 741

southwest Wales, 1846-1947. Ph.D. Thesis, Aberystwyth University, Aberystwyth, Wales. 742

743

Jones, M.R., Fowler, H.J., Kilsby, C.G., Blenkinsop, S., 2012a. An assessment of changes in 744

seasonal and annual extreme rainfall in the UK between 1961 and 2009. International 745

Journal of Climatology 33, 1178-1194. 746

747

Jones, P.D., Harpham, C., Briffa, K.R., 2012b. Lamb weather types derived from reanalysis 748

products. International Journal of Climatology 33, 1129-1139. 749

750

Kehew, A.E., Milewski, A., Soliman, F., 2010. Reconstructing an extreme flood from boulder 751

transport and rainfall–runoff modelling, Wadi Isla, South Sinai, Egypt. Global and Planetary 752

Change 70, 64-75. 753

754

Lamb, H.H., 1972. British Isles Weather Types and a Register of the Daily Sequence of 755

Circulation Patterns. Geophysical Memoirs 116, HMSO, London. 756

757

Lewin, J., Macklin, M.G., 2003. Preservation potential for Late Quaternary river alluvium. 758

Journal of Quaternary Science 18, 107-120. 759

Linderholm, H.W., Folland, C.K., Hurrell, J.W., 2008. Reconstructing summer North Atlantic 760

Oscillation (SNAO) variability over the last five centuries. In: Elferts, D., Brumelis, G., 761

Gärtner, H., Helle, G., Schleser, G. (Eds.), TRACE - Tree Rings in Archaeology, Climatology 762

31

and Ecology. Proceedings of the DENDROSYMPOSIUM 2007. Scientific Technical Report 763

08/05, Vol. 6, Potsdam, Germany, pp. 6-13. 764

Longfield, S.A., Macklin, M.G., 1999. The influence of recent environmental change on 765

flooding and sediment fluxes in the Yorkshire Ouse basin. Hydrological Processes 13, 1051-766

1066. 767

Lumbroso, D., Gaume, E., 2012. Reducing the uncertainty in indirect estimates of extreme 768

flash flood discharges. Journal of Hydrology 414, 16-30. 769

770

Maas, G., Macklin, M.G., 2002. The impact of recent climate change on flooding and 771

sediment supply within a Mediterranean mountain catchment, southwestern Crete, Greece. 772

Earth Surface Processes and Landforms 27, 1087-1105. 773

774

Macdonald, N., 2012. Trends in flood seasonality of the River Ouse (northern England) from 775

archive and instrumental sources since AD 1600. Climatic Change 110, 901-923. 776

777

Macdonald, N., 2013. Reassessing flood frequency for the River Trent through the inclusion 778

of historical flood information since AD 1320. Hydrology Research, 779

DOI:10.2166/nh.2012.188. 780

781

Macdonald, N., Black, A.R., 2010. Reassessment of flood frequency using historical 782

information for the River Ouse at York, UK (1200–2000). Hydrological Sciences Journal 55, 783

1152-1162. 784

785

Macdonald, N., Jones, C.A., Davies, S.J., Charnell‐White, C., 2010. Historical weather 786

accounts from Wales: an assessment of their potential for reconstructing climate. Weather 787

65 (3), 72-81. 788

32

Macklin, M.G., Rumsby, B.T., 2007. Changing climate and extreme floods in the British 789

uplands. Transactions of the Institute of British Geographers 32, 168-186. 790

791

Macklin, M.G., Rumsby, B.T., Heap, T., 1992. Flood alluviation and entrenchment, Holocene 792

valley floor development and transformation in the British uplands. Geological Society of 793

America Bulletin 104, 631-643. 794

795

Marchi, L., Borga, M., Preciso, E., Gaume, E., 2010. Characterisation of selected extreme 796

flash floods in Europe and implications for flood risk management. Hydrological Processes 797

394, 118-133. 798

799

McEwen, L.J., 2006. Flood seasonality and generating conditions in the Tay catchment, 800

Scotland from 1200 to present. Area 38, 47-64. 801

802

McEwen, L.J., Werritty, A., 2007. ‘The Muckle Spate of 1829’: the physical and societal 803

impact of a catastrophic flood on the River Findhorn, Scottish Highlands. Transactions of the 804

Institute of British Geographers 32 (1), 66-89. 805

806

Merrett, S.P, Macklin, M.G., 1999. Historic river response to extreme flooding in the 807

Yorkshire Dales, northern England. In: Brown, A.G., Quine, T.M. (Eds.), Fluvial Processes 808

and Environmental Change. Wiley, Chichester, UK, pp. 345-360. 809

810

Milan, D., 2012. Geomorphic impact and system recovery following an extreme flood in an 811

upland stream, Thinhope Burn, northern England, UK. Geomorphology 138, 319-328. 812

813

Newson, M., 1975. The Plynlimon Floods of August 5th/6th 1973. Natural Environment 814

Research Council Report no. 26, Institute of Hydrology, Wallingford, UK. 815

33

Newson, M., 1980. The geomorphological effectiveness of floods – a contribution stimulated 816

by two recent events in mid-Wales. Earth Surface Processes and Landforms 5, 1-16. 817

818

Newson, M.D., 1989. Flood effectiveness in river basins, progress in Britain in a decade of 819

drought. In: Beven, K., Carling., P.A. (Eds.), Floods, Hydrological, Sedimentological and 820

Geomorphological Implications. Wiley, Chichester, UK, pp. 151-169. 821

822

Osborn, T.J., 2006. Recent variations in the winter North Atlantic Oscillation. Weather 61, 823

353-355. 824

825

Osborn, T.J, Hulme, M., 2002. Evidence for trends in heavy rainfall events over the UK. 826

Philosophical Transactions of the Royal Society of London, Series A 360, 1313-1325. 827

828

Overland, J.E., Francis, J.A., Hanna, E., Wang, M., 2012. The recent shift in early summer 829

Arctic atmospheric circulation. Geophysical Research Letters 39 (19), L19804. 830

DOI:10.1029/2012GL053268. 831

832

Parry, L.l., 2013. Documenting memories of extreme weather and flooding in north 833

Ceredigion, West Wales. Unpublished M.Sc. thesis, Aberystwyth University, Aberystwyth, 834

Wales, 77 pp. 835

836

Pattison, I., Lane, S.N., 2012. The relationship between Lamb weather types and long-term 837

changes in flood frequency, River Eden, UK. International Journal of Climatology 32, 1971-838

1981. 839

840

Roca, M., Davison, M., 2009. Two dimensional model analysis of flash-flood processes, 841

application to the Boscastle event. Journal of Flood Risk Management 3, 63-71. 842

843

34

Rodda, H.J., Little, M.A., Wood, R.G., MacDougall, N., McSharry, P.E., 2009. A digital 844

archive of extreme rainfalls in the British Isles from 1866 to 1968 based on British Rainfall. 845

Weather 64 (3), 71-75. 846

847

Ruiz-Villanueva, V., Díez-Herrero, A., Bodoque, J.M., Ballesteros Cánovas, J.A., Stoffel, M., 848

2013. Characterisation of flash floods in small ungauged mountain basins of central Spain 849

using an integrated approach. Catena 110, 32-43. 850

851

Rumsby, B.T. 1991. Flood frequency and magnitude estimates based on valley floor 852

morphology and floodplain sedimentary sequences: the Tyne Basin, NE England. 853

Unpublished Ph.D. thesis, Newcastle University, Newcastle, UK, 208 pp. 854

855

Rumsby, B.T., Macklin, M.G., 1994. Channel and floodplain response to recent abrupt 856

climate change, the Tyne basin, northern England. Earth Surface Processes and Landforms 857

19, 499-515. 858

859

Rumsby, B.T., Macklin, M.G., 1996. River response to the last Neoglacial cycle (the ‘Little 860

Ice Age’) in northern, western and central Europe. In: Branson, J., Brown, A.G., Gregory, 861

K.J. (Eds.), Global Continental Changes, The Context of Palaeohydrology. Geological 862

Society Special Publication 115, London, UK, pp. 217-233. 863

864

Screen, J.A., 2013. Influence of Arctic sea ice on European summer precipitation. 865

Environmental Research Letters 8 (4), 044015. DOI: 10.1088/1748-9326/8/4/044015. 866

867

Screen, J.A., Simmonds, I., 2013. Exploring links between Arctic amplification and mid‐868

latitude weather. Geophysical Research Letters 40 (5), 959-964. 869

870

35

Smith, A., Bates, P., Freer, J., Wetherhall, F., 2013. Investigating the application of climate 871

models in flood projection across the UK. Hydrological Processes, DOI: 10.1002/hyp.9815. 872

873

Uhlemann, S., Bertelmann, R., Merz, B., 2013. Data expansion: the potential of grey 874

literature for understanding floods. Hydrology and Earth System Sciences 17, 895-911. 875

Walsh, S., 2010. Report of Rainfall of November 2009. Met Éireann, Climatological Note 12, 876

pp.1-17. 877

878

Wells, S.G., Harvey, A.M., 1987. Sedimentologic and geomorphic variations in storm-879

generated alluvial fans, Howgill Fells, northwest England. Geological Society of America 880

Bulletin 98, 182-198. 881

882

Wilby, R.L., Quinn, N.W., 2013. Reconstructing multi-decadal variations in fluvial flood risk 883

using atmospheric circulation patterns. Journal of Hydrology 487, 109-121. 884

885

Wilby, R.L., Beven, K.J., Reynard, N.S., 2008. Climate change and fluvial flood risk in the 886

UK, more of the same? Hydrological Processes 22, 2511-2523. 887

888

Winchester, V., Chaujar, R.K., 2002. Lichenometric dating of slope movements, Nant 889

Ffrancon, north Wales. Geomorphology 47, 61-74. 890

891

Newspapers and archive material 892

893

Aberystwyth Observer. 1870. Great floods at Aberystwyth. Aberystwyth Observer, 5th 894

November 1870. 895

896

Aberystwyth Observer. 1878a. No title. Aberystwyth Observer, 16th November 1878. 897

898

Aberystwyth Observer. 1878b. Aberaeron. Aberystwyth Observer, 16th November 1878. 899

36

Aberystwyth Observer. 1879. Llanbadarn Fawr: Tremendous flood in the Rheidol. 900

Aberystwyth Observer, 23rd August 1879. 901

902

Aberystwyth Observer. 1880. The floods. Aberystwyth Observer, 25th December 1880. 903

904

Aberystwyth Observer. 1881a. Aberaeron flood – town bridge. Aberystwyth Observer, 1st 905

January 1880. 906

907

Aberystwyth Observer. 1881b. The floods. Aberystwyth Observer, 15th January 1881. 908

909

Aberystwyth Observer. 1886a. Great floods. Aberystwyth Observer, 23rd October 1886. 910

911

Aberystwyth Observer. 1886b. Tregaron. Aberystwyth Observer, 25th December 1886. 912

913

Aberystwyth Observer. 1889. Local and district news - Floods. Aberystwyth Observer, 30th 914

March 1889. 915

916

Aberystwyth Observer. 1898. Heavy thunderstorm. Aberystwyth Observer, 16th June 1898. 917

918

Aberystwyth Observer. 1894. Heavy floods. Aberystwyth Observer, 15th November 1894. 919

920

Aberystwyth Observer. 1903. Heavy flood. Aberystwyth Observer, 10th September 1903. 921

922

Aberystwyth Observer. 1909. River Rheidol in flood – exciting scenes. Aberystwyth 923

Observer, 22nd July 1909. 924

925

Aberystwyth Observer. 1910. Remarkable rainfall. Aberystwyth Observer, 8th December 926

1910. 927

928

Cambrian. 1846. Destructive flood. Cambrian, 7th August 1846. 929

930

Cambrian News. 1869. Great floods in Wales. Cambrian News (Aberystwyth edition), 13th 931

February 1869. 932

933

Cambrian News. 1886a. The great storm and flood at Aberystwyth. Cambrian News 934

(Aberystwyth edition), 22nd October 1886. 935

936

37

Cambrian News. 1886b. Tregaron. Cambrian News (Aberystwyth edition), 25th December 937

1886. 938

939

Cambrian News. 1870. The Floods. Cambrian News (Aberystwyth edition), 29th October 940

1870. 941

Cambrian News. 1909. River Rheidol in flood: Remarkable scenes near Aberystwyth. 942

Cambrian News (Aberystwyth edition), 24th July 1909. 943

944

Cambrian News. 1910a. Prolonged thunderstorm. Cambrian News (Aberystwyth edition), 945

10th June 1910. 946

947

Cambrian News. 1910b. Tregaron - thunderstorm. Cambrian News (Aberystwyth edition), 948

16th December 1910. 949

950

Cambrian News. 1911. Tregaron - floods. Cambrian News (Aberystwyth edition), 17th 951

November 1911. 952

953

Cambrian News. 1919. A sudden storm. Cambrian News (Aberystwyth edition), 20th June 954

1919. 955

956

Cambrian News. 1922. Fighting the flood at Tregaron. Cambrian News (Aberystwyth 957

edition), 22nd September 1922. 958

959

Cambrian News. 1947. The blizzard’s aftermath in Cardiganshire. Cambrian News 960

(Aberystwyth edition), 21st March 1947 (also 28th March edition). 961

962

Cambrian News. 1957. Feeding in the flood. Cambrian News (Aberystwyth edition), 9th 963

August 1957. 964

965

Cambrian News. 1964. And now to count the cost, cloudburst Saturday and Wales 966

overflows. Cambrian News (Aberystwyth edition), 18th December 1964. 967

968

Cambrian News. 1973. The washout weekend. Cambrian News (Aberystwyth edition), 10th 969

August 1973. 970

971

Cambrian News. 1987. Worst floods since 1960. Cambrian News (Aberystwyth edition), 972

23rd October 1987. 973

38

Cambrian News. 1998. Rain stops play over weekend. Cambrian News (Aberystwyth 974

edition), 12th March 1998. 975

976

Cambrian News. 2000. Batten down the hatches. Cambrian News (Aberystwyth edition), 2nd 977

November 2000. 978

979

Cambrian News. 2012. Mayhem in wake of worst flooding in living memory. Cambrian News 980

(Aberystwyth edition), 14th June 2012. 981

982

Cwmystwyth School Logbook. 1921. Ceredigion Archives, Aberystwyth Library, Aberystwyth. 983

984

Dydd. 1886. Llifogydd Mawrion. Dydd (Dolgellau), 17th September 1886. 985

986

Elerch School Logbooks. 1909, 1924, 1926, 1929, 1930. Ceredigion Archives, Aberystwyth 987

Library, Aberystwyth. 988

989

Gwyliedydd. 1881. Llifogydd Mawrion yn Aberystwyth. Y Gwyliedydd (Rhyl), 5th January 990

1881. 991

992

Gwyliedydd. 1886. No title. Y Gwyliedydd (Rhyl), 20th October 1886. 993

994

Llanafan School Log Books. 1936, 1939. Ceredigion Archives, Aberystwyth Library, 995

Aberystwyth. 996

997

Llanfihangel-y-Creuddyn School Log Book. 1919. Ceredigion Archives, Aberystwyth Library, 998

Aberystwyth. 999

1000

Llanilar School Log Books. 1903, 1910, 1940. Ceredigion Archives, Aberystwyth Library, 1001

Aberystwyth. 1002

1003

Montgomeryshire Express and Radnor Times. 1909. Hail and hurricane – floods in the 1004

Severn. Montgomeryshire Express and Radnor Times, 7th December 1909. 1005

1006

Papur Pawb. 1979. Papur Pawb 49, June 1979. 1007

1008

Papur Pawb. 1987. Papur Pawb 125, January 1987. 1009

1010

39

Seren Cymru, 1852. Hanesion Cartrefol. Seren Cymru (Carmarthen), Thursday 25th 1011

November 1852. 1012

1013

Seren Cymru. 1877. Y Llifogydd diweddar. Seren Cymru (Carmarthen), 7th September 1877. 1014

1015

Seren Cymru. 1879. Ystorom o fellt a tharanau. Seren Cymru (Carmarthen), 22nd August 1016

1879. 1017

1018

The Examiner. 1842. Flood at Aberystwith, 24th September 1842. 1019

1020

The Times .1852. The Inundations. The Times (London), Thursday 15th November 1852, 1021

Issue 21273, p.8. (Also issues 21274, 21276, 21279, on 16th, 18th and 22nd November). 1022

1023

The Times. 1922. Floods in Wales. The Times (London), Wednesday 20th September 1922, 1024

issue 43141, p.10. 1025

1026

The Times. 1931. Floods in Wales. The Times (London), Monday 15th June 1931, p. 12. 1027

1028

The Times. 1948. Flooding after heavy rain. The Times (London), Monday 12th January 1029

1948, Issue 50966, p.4. 1030

1031

The Times. 1957. Further damage by gales and floods. The Times (London), Thursday 26th 1032

September 1957, Issue 53956, p.6. 1033

1034

Welsh Gazette. 1919. Borth, the storm. Welsh Gazette, June 1919. 1035 1036 1037 Welsh Gazette. 1935. Strom havoc in Cardiganshire. Welsh Gazette, 27th June 1935. 1038

1039

Y Cymro. 1964. Y Cymro, 17th December, 1964. 1040

1041

1042

40

Figure captions 1043

Fig. 1. The upper Ystwyth study area showing locations of the main headwater 1044

tributaries where boulder berms have been dated (numbered 1-3). Graveyard 1045

locations used to construct lichen size–age relationships (Hafod and Ystbyty 1046

Ystwyth) are also shown. Inset map (i) shows the neighbouring upper Rheidol and 1047

upper Leri catchments, where a small number of boulder berms and debris cones 1048

were identified; this expands to inset map (ii), which shows the wider UK context. 1049

1050

Fig. 2. Examples of geomorphological features dated in the Cambrian Mountains, 1051

including boulder berms in the Nant Cwm-Du (2A) and Nant Gau catchments (2B, 1052

2C), and boulder deposits associated with a steep hillslope gully (2D). 1053

1054

Fig. 3. (A) Size–age relationships for Porpidia tuberculosa and Rhizocarpon 1055

geographicum in the Cambrian Mountains; (B) differences between predicted and 1056

inscribed gravestone ages (n = 20) based on the size-age regression equation for P. 1057

tuberculosa. For the 3LL method there was no correlation between lichen size and 1058

dating error. 1059

1060

Fig. 4. Summary decadal frequency plots of boulder-berm, fan and cone 1061

sedimentation in the Cambrian Mountains. Error bars show two standard deviations 1062

based on gravestone age tests (Table 1). It is important to note that low ‘apparent’ 1063

rates of berm deposition before 1900 reflect a combination of undateable surfaces 1064

41

(because of coalescing lichens) and reworking of older deposits by more recent 1065

floods. 1066

1067

Fig. 5. View upstream through the middle reaches on the Nant Cwm-Du catchment. 1068

In this small system much of the berm material appears to have been sourced from 1069

steep and unstable hillslopes, which show evidence of historical erosion, although 1070

they are now relatively stable and vegetated. 1071

1072

Fig. 6. Upper Leri catchment showing the active channel (cobble/boulder bed), 1073

floodplain and boulders deposited in June 2012 (bag for scale is 0.4 x 0.3 m) and a 1074

low terrace and boulder-berm lichen dated to 1932 ±10. At this, point the Leri has a 1075

drainage area of 7 km2 and an average channel bed slope of 0.015 m m-1. 1076

1077

Fig. 7. Cumulative seasonal (DJF, MAM, JJA, SON) plots of documentary floods 1078

reported in Table 2 and overlaid with lichen dates for all floods (channel and slope). 1079

1080

Fig. 8. Relative flood magnitude based on average B-axis measurements (Di) of the 1081

five largest clasts present in each boulder-berm. 1082

1083

Fig. 9. Nant Cwm-Du discharge estimates based on Carling (1986) with roughness 1084

derived from Jarrett (1992) plotted alongside European and UK extreme flood 1085

estimates for catchments of < 50 km2 drainage area. 1086

42

Fig. 10. Correlations between July-August rainfall (A), average and maximum 1087

gauged daily flows (B, C), and the SNAO index of Folland et al. (2009) at 1088

Cwmystwyth. Rainfall data cover the period 1961-2010 and flow data 1984-2011. 1089

Coefficients of correlation (r) are also shown where single and double asterisks 1090

indicate significance at p = <0.05 and p = <0.01, respectively. 1091

1092

Fig. 11. Summary decadal frequency plots of all dated geomorphological features in 1093

the Cambrian Mountains plotted against (A) July-August SNAO index; Folland et al., 1094

2009); (B) July-August frequency of cyclonic LWTs and smoothed July-August 1095

rainfall in Southwest England and Wales (SWEP), based on Alexander and Jones 1096

(2001), and (C) average November-February NAO index and smoothed total 1097

November-February rainfall (SWEP). Low ‘apparent’ rates of berm deposition before 1098

1900 reflect a combination of undateable surfaces (owing to coalescing lichens) and 1099

reworking of older deposits by more recent floods. 1100

1101

Fig. 12. Decadal frequency plots of all lichen-dated boulder berms (channel floods) in 1102

Wales and England. The dashed line indicates low data quality/no data associated 1103

with berms that are either vegetated or beyond the age range of lichenometry 1104

(typically very large coalesced specimens). The original North Pennines plot 1105

(Macklin et al., 1992) has been updated with data from an extreme flood in 2007 1106

(Milan, 2012). 1107

Table 1

Accuracy statistics of LL (largest lichen), 3LL, and 5LL (mean of the 3 and 5 largest specimens) lichen size–age curves for Porpidia tuberculosa based on surface age predictions of gravestones

Statistic Maximum

lichen

Mean of 3 largest

lichen

Mean of 5 largest

lichen

Maximum accuracy (yrs) Exact year Exact year Exact year

Minimum accuracy (yrs) 32 20 19

Mean accuracy (yrs) 09 09 10

Standard deviation (yrs) 08 05 05

% Predictions > 5 years

accuracy

40 20 20

% Predictions > 10 years

accuracy

55 65 55

% Predictions > 15 years

accuracy

85 90 90

Table 2

Documentary flood events in the Cambrian Mountains, 1842-2012a

Date River / catchment / area

Rainfall Flooding Reference

Sep 1842 Rheidol - ‘..a noise was heard resembling that of distant thunder…a tremendous body of water was seen rolling several feet above the waters of the River Rhydol, stripping the fields of hay, wheat, oats and barley.’

The Examiner (1842)

30th Jul 1846

Rivers draining Mynydd Bach – Aeron, Arth, Peris, Cledan

Heavy & localised rainfall over Mynydd Bach area of south Ceredigion

‘One of the most dreadful floods that ever occurred in the principality took place on Thursday night in Cardiganshire.’ Bridges destroyed & two known fatalities.

Cambrian (1846), Jones (2013)

Mid-Nov 1852 Ystwyth, Rheidol ‘Y mae cymaint o wlaw wedi disgyn yn y mis a aeth heibio…’ (so much rain has fallen in this last month).

‘the Ystwyth and Rheidol rivers have overflowed their banks and produced great destruction…’

The Times (1852), Seren Cymru (1852)

7th/8th Feb 1869 Severn, Vyrnwy, Dyfi & southern Snowdonia area.

‘a strong wind from the SW set in and continued during the whole of Sunday, accompanied with an almost incessant rainfall.’

‘For a distance of 7-8 miles there was nothing visible but one great sheet of water, which submerged the lowlands to a depth of 7-8 feet, carrying away in its course a great quantity of wreck.’

Cambrian News (1869)

30th/31st Oct, Ystwyth, Rheidol, ‘unusually heavy rain…’ ‘The depth of water was in some parts between 5-6 feet…Several wooden

Cambrian News (1870), Aberystwyth

a Newspaper and archival references are given in a separate bibliography at the end of this article. British Rainfall entries can be accessed via a digital archive (www.badc.nerc.ac.uk; Rodda et al., 2009). Where references to specific properties are made in school log books, these have been anonymised with [blank].

1870 Severn bridges were carried away’. ‘Such a flood has not been witnessed in this town during the memory of that often quoted individual “the oldest inhabitant”.’

Observer (1870)

Late Jan 1876 Ystwyth - ‘Attendance fell off this week owing to the unusual severity of the weather. The children cannot attend as the rivers are continually flooded.’

Cwmystwyth School Log Book (1876)

Late Aug/early Sep 1877

Teifi, Wye, Tywi - ‘The extensive damage to property wrought in recent weeks by the large floods in the south call for attention and sympathy throughout the country.’

Seren Cymru (1877)

10th Nov 1878 Rheidol, Ystwyth, Aeron

- ‘The heavy floods on Sunday carried away two wooden foot-bridges on the Ystwyth and Aeron. The Rheidol, and all other Welsh rivers, were swollen and overflowed their banks.’

Aberystwyth Observer (1878a, b)

16th/17th Aug 1879

Ystwyth, Rheidol, Aeron, Teifi

‘a storm of unprecedented severity’ ‘for nearly 24 hours the rain did not cease falling.’ Widespread totals of 60-75 mm.

‘..the greatest flood here within the memory of the present generation. It exceeded the flood at the beginning of November last…’

Aberystwyth Observer (1879), Seren Cymru (1879), British Rainfall

22nd Dec 1880 Ystwyth, Rheidol, Aeron

‘Heavy rains had fallen for two or three days previously.’

‘The "oldest inhabitant" of Aberystwyth will have his memory seriously taxed to recall a flood of such magnitude as that which visited the town on Wednesday evening.’ Reports of bridges on Ystwyth and Rheidol destroyed.

Aberystwyth Observer (1880, 1881a, b), Gwyliedydd (1881)

10th Sep, 1886 Brenig - ‘Residents awoke terrified by the noise of water entering their homes. In a short time the whole of Tregaron was covered by water, measuring 4-5 feet in depth.’

Dydd (1886)

15/16th Oct 1886 Rheidol, Ystwyth, Teifi, Leri, Dyfi

Daily total of 97 mm at Cwmsymlog

The 'Great Storm' of October 1886. ‘Trefechan bridge has stood the test of many storms of wind and flood, but that of Saturday morning proved one too many for it.’ Felin Newydd bridge also destroyed.

Cambrian News (1886a), Aberystwyth Observer (1886a)

21st Dec 1886 Brenig - The thaw of snow which set in swelled the River Brenig to such an extent that it overflowed, covering the whole town with a great depth of water.

Aberystwyth Observer (1886b)

25th – 26th Mar 1889

Ystwyth, Rheidol ‘Rain must have fallen heavily on the hills on Saturday and Sunday…’

‘….the rivers Ystwyth and Rheidol were greatly flooded on the following days.’

Aberystwyth Observer (1889)

15th Nov 1894 Ystwyth 50-75 mm of rain over parts of the Cambrian Mountains.

‘Morning train out of Aberystwyth unable to proceed over Llanilar flats. Traffic resumed later that morning.’

Aberystwyth Observer (1894), British Rainfall

11th-12th Jun 1898

Tregaron area ‘..great heat prevailed…the sky became overcast…vivid flashes of lightning and peals of thunder… the rainfall was very heavy in many districts.’

Heavy floods between Tregaron and Derry Ormond.

Aberystwyth Observer (1898)

5th Jan 1903 Ystwyth - ‘Raining very heavily; great floods in the rivers so that the number of children present was only 64.’

Llanilar School Log Book (1903)

10th Sep1903 Rheidol (and Ystwyth)

‘The rain which fell almost continuously from Monday afternoon until early this morning…’

‘…resulted in the flooding of local rivers. Soon after 5 am the Rheidol valley was flooded and the whole of the meadows were underwater by 8 am; waters had subsided by mid-day.’

Aberystwyth Observer (1903)

15th Jul 1909 Rheidol, Leri - ‘..It is stated that this was the largest flood since the bridges were carried away 18 years ago, but others say there was a similar flood 12 years ago..’

Cambrian News (1909), Aberystwyth Observer (1909), Elerch School Log Book (1909)

28th Sep, 1909 Leri - ‘Two children from [blank] have been absent since Tuesday, owing to the river being swollen.’

Elerch School Log Book (1909)

2nd Dec 1909 Severn ‘On Wednesday and Thursday torrential rain fell over Newtown and the neighbourhood.’

‘On Thursday morning the downpour in the upper Severn valleys caused the river to rise rapidly; in the early hours of Friday morning it had risen 6-7 feet above its normal level.’

Montgomeryshire Express and Radnor Times (1909)

11th Feb 1910 Ystwyth - ‘Attendance for the week has been affected by a flood in the river.’

Llanilar School Log Book (1910)

7th Jun 1910 Aberystwyth, Tregaron, Llangurig

‘The wind switched to the southeast; hot, stale air from Russia & Germany flooded the country. Torrents of rain fell – ca. 45 mm in Aberystwyth. Over the Plynlimon range totals were probably in the order of >50 mm in 12 hours.’

Severe thunderstorms; ‘much flooding has been caused and roads have been rendered impassable.’ Great floods experienced on the Teifi.

Cambrian News (1910a)

7th Dec1910 Severn ‘Rain continued to fall, until the total amount for the 36 hours ending 6 pm was 55 mm.’

‘Such a great weight of rain as recorded for Thursday has naturally led to extensive floods.’

Aberystwyth Observer (1910)

12th Dec1910 Brenig ‘A sudden storm passed over the district on Monday night. Flashes of lightning were followed by heavy peals of thunder. Heavy showers of hail and rain fell afterwards….

‘… the River Brenig flooded. There was a similar storm on Tuesday night.’

Cambrian News (1910b)

13th Nov 1911 Rheidol, Teifi - After an Indian summer the weather changed in November. The Rheidol valley was inundated and the main road Pwllhobi was flooded. Section of the light railway washed away.

Cambrian News (1911)

10th Jun 1913 Ystwyth - ‘Last night and this morning witnessed very heavy rain and floods in the district; very few children present.’

Cwmystwyth School Log Book (1913)

12th Mar 1919 Ystwyth - ‘Owing to stormy weather and floods only 21 children attended school.’

Llanfihangel School Log Book (1919)

12th Jun 1919 Rheidol, Leri 91 mm at Strata Florida, 104 mm at Devil's Bridge, 75 mm at Gogerddan & 87 mm at Tal-y-bont. ‘..a severe westerly gale, unusual and unexpected in June..’

‘The Leri overflowed, causing much damage.’ ‘..the Rheidol overflowed, flooding adjacent meadows.’

Cambrian News (1919), Welsh Gazette (1919)

3rd Nov 1921 Cwmystwyth - “Attendance is very good this week with the exception of [blank], who experience great trouble from

Cwmystwyth School Log Book (1921)

dangerous floods on the mountain side. On Thursday 3rd of this month, their father had to come to their assistance before they could get home.”

19th Sep 1922 Rheidol, Severn ‘A large and deep cyclonic disturbance.’ 36 mm of rain at Aberystwyth; heavy rain and floods reported in west and north Wales.

The Rheidol overflowed into the streets of Aberystwyth and a big landslide was reported on Plynlimon; a bridge over the Rheidol at Capel Bangor collapsed. ‘The country around Strata Florida was a scene of devastation.’ One fatality near Machynlleth.

The Times (1922), Cambrian News (1922)

2nd Jun1924 Leri ‘Exceptionally heavy rain.’ ‘One of the bridges across the river is not very safe.’

Elerch School Log Book (1924)

21st Jul 1924 Leri ‘An exceptionally wet morning again; it has rained heavily in the night and continues to do so.’

‘17 children present in the morning, 21 in the afternoon.’

Elerch School Log Book (1924)

18th Jul 1926

Leri ‘one of the worst thunderstorms in living memory’. 75-100 mm over Cambrian Mountains east of Aberystwyth.

‘The river is flooded…’ Elerch School Log Book (1926), British Rainfall

6th Sep 1927 Leri 0.85 inches (22 mm), Aberystwyth

- Interview with Tal-y-bont resident (Parry, 2013)

14th/15th Nov, 1929

Leri, Ystwyth - 'The floods have washed away the bridge from the river near [blank]…’

Elerch School Log Book (1929)

28th Mar, 1930 Leri - ‘The children of [blank] cannot come to school because the bridge near their house has been washed away…’

Elerch School Log Book (1930)

14th Jun 1931 Severn Mid-Wales was this afternoon visited by a great thunderstorm which caused serious and widespread havoc. In less than two hours rain had fallen in such a deluge…’

Roads flooded and pavements damaged.

The Times (1931)

25th Jun 1935 Rheidol, Leri ‘On 25th June a depression over the Bay of Biscay moved northwards accompanied by heavy rain/thunder. Totals of 70 mm & 83 mm from 15.00 to 20.00.

‘A terrible thunderstorm accompanied by torrential rain and lightning– river flooded and bridges swept away.’

British Rainfall; Welsh Gazette (1935) Cambrian News (1935), Elerch School Log Book (1935)

14th Dec 1936 Ystwyth - ‘Very inclement weather – wet, stormy – floods. Children present = 11/46.’

Llanafan School Log Book (1936)

31st Jul 1939 Ystwyth - ‘Most inclement weather and in many cases the river had flooded so that several pupils could not attend.’

Llanafan School Log Book (1939)

4th Nov1940 Ystwyth - ‘Heavy rain has caused the Ystwyth to overflow and the 3 pupils who live on the other side of the river cannot attend.’

Llanilar School Log Book (1940)

18th Mar 1947 Ceredigion & Powys; Dyfi, Teifi

Severe blizzards and deep snow drifts in mid-March gave way to milder air and

‘Due to the rapid thaw and heavy rainfall, hundreds of acres of farmland have been inundated. In Cardiganshire, considerable flooding

Cambrian News (1947)

heavy rainfall. resulted.’

11th Jan 1948 Teifi, Severn, Vyrnwy

Severe floods occurred after heavy rain.

Inhabitants of the lower Teifi valley were forced to move their belongings to higher ground.

The Times (1948)

26thJul 1957 Rheidol - ‘The thunderstorm which swooped on Cardiganshire left a trail of havoc. Floodwater 5 feet deep in places. A landslide blocked the old coach road between Cwmystwyth and Rhayader.’

Cambrian News (1957)

5th Aug 1957 Cambrian Mountains Thunderstorms & 88 mm of rain at Gwngu. 102 mm in 2 hours near Llangurig.

‘Sheep and cattle washed away and drowned as streams rose rapidly following exceptionally heavy rain’.

British Rainfall, Newson (1975)

25th Sep 1957 Cambrian Mountains 38 mm at Aberystwyth ‘Llandrindod Wells was cut off by floods and three landslides last night…after the heavy rains the mountain streams were swollen to many times their usual size’.

The Times (1957)

12th Dec 1964 Ystwyth, Rheidol, Leri, Dyfi, Tywi, Conwy

122 mm at Cwmystwyth A ‘night of terror’ in the Ystwyth valley. Bridges and roads washed away and several large landslides. Weir at Tal-y-bont destroyed.

Cambrian News (1964), Y Cymro (1964)

5th Aug 1973 Rheidol, Clarach ‘Three and a half inches fell in two days’ in Aberystwyth.

Railway embankments and road bridges damaged.

Cambrian News (1973); Newson (1980)

15th Aug 1977 Upper Severn & Tallwyth

Highly localised thunderstorm - 90 mm in 2 hours.

Highly localised flood. Redistribution of gravel and cobble shoals.

Newson (1980)

11th May 1979 Clarach, Peithyll, Ceulan, Leri

‘…heavy rain fell in the Aberystwyth area all day and

- Y Tincer (1979), Papur Pawb (1979)

night on 10th…’

5th Dec 1979 Aeron - ‘…roads were flooded in Dyfed, and in Aberayron four feet of water swamped houses.’

The Times (1979)

18th Oct1987 Aeron - Worst flood since 1960 Cambrian News (1987)

6th Mar 1998 Rheidol, Ystwyth, Clarach, Dyfi, Severn

- ‘…trains in and out of Aberystwyth were cancelled…’

Cambrian News (1998)

30th Oct 2000 Ceulan, Rheidol - Although several roads flooded in Aberystwyth, flows on the lower Leri/upper Wye were 3.5 to 5.0 times lower than maximum recorded values.

Cambrian News (2000)

8th June 2012 Leri, Rheidol, Clarach

Maximum daily fall of 146 mm in the Rheidol catchment. 160 mm in 24 hours on Pumlumon.

Widespread and locally severe flooding in west Wales. Leri, Rheidol and Clarach worst affected.

Cambrian News (2012); Foulds et al., (2012)

Table 3

Discharge estimates for 29 boulder berms in the Nant Cwm-Du catchment based on Costa (1983), Carling (1986), and a variety on n values

Method Q mean

(m3 s-1)

Q range

(m3 s-1)

(1) Carling (1986); n = 0.05 9 ± 4 4 – 19

(2) Carling (1986); n = 0.07 7 ± 3 3 – 14

(3) Carling (1986); n = 0.32 S0.38 R-0.16 (Jarrett, 1992) 3 ± 1 1 – 7

Mean of methods 1, 2 & 3 6 ± 3 3 – 13

(4) Costa (1983); Q = VA (V = 0.18 Di 0.487) 33 ± 17 10 – 73

Table 4

Flood discharges in small UK catchments compared to estimated values in the Nant Cwm-Du catchment using a combination of Carling (1986) (Q), and Jarrett (1992) (n)

Flood location

and date

Drainage area

(km2)

Q (m3 s-1) QU (m3 s-1 km2)

Reference

West Stream, Dorset, May 1986

0.80 7 8.8 Acreman (1989)

Afon Tanllwyth, Cambrian

Mountains, Aug 1973 0.89 2.16 2.4 Newson (1975)

Nant Cwm-Du,

Cambrian Mountains 0.90 1-7 1.1–7.8 This study

Nant Iago, Cambrian Mountains, Aug 1973

1.02 5.65 5.5 Newson (1975)

Raise Beck, Lake District,

Jan 1995 1.27 4–6 3.2–4.7 Johnson and

Warburton (2002)

Jordan River, Boscastle, Aug 2004

2.30 20 8.7 Roca and Davison (2009)

West Grain, North Pennines, Jul 1983

3.92 16-22 4.1–5.6 Carling (1986)

Ireshope Burn, North Pennines, Jul 1983

5.93 35 5.9 Carling (1986)

Langdon Beck, North Pennines, Jul 1983

7.14 30 4.2 Carling (1986)

Upper Wye, Cambrian Mountains, Aug 1956

10.6 69 6.5 Environment Agency (2013)

Thinhope Burn, North Pennines, Jul 2007

12.0 60 5.0 Milan (2012)

Table 5

Documentary flood dates (taken from Table 2) and corresponding NAO/SNAO index and LWT values for three days before and on the day of flood (the latter indicated by bold values)

Flood date NAO LWT Sep 1842 - -

30/08/1846 - - 15/11/1852 -0.93 - 7-8/02/1869 3.90 -

30-31/10/1870 1.3 - Jan 1876 1.1 -

Aug/Sep 1877 -0.20/-3.74 - 10/11/1878 -3.47 N, CN, NW, C 16/08/1879 -0.54, 0.43, 0.62, -0.24 SW, A, CNE, C 22/12/1880 0.52 C,C,A,CS 10/09/1886 - SW, SW, SW, W 15/10/1886 -0.55 C, CNW, W, C 21/12/1886 0.12 N, CNE, A, SW 25/03/1889 0.11 A,A,A,AW 15/11/1894 1.96 SW, S, C, S 11/06/1898 - ASE, A, AE, AE 05/01/1903 1.28 C, CW, W, S 10/09/1903 -0.27 C, A,C, C 15/07/1909 -1.15, -0.91, -0.56, -0.56 ANW, W, W, W 28/09/1909 -0.52 AE, A, ASE, SE 02/12/1909 -0.48 CSW, C, C, C 11/02/1910 3.85 N, A, SW, W 07/06/1910 -0.38 CE, E, E, E 13/11/1911 -0.05 N, CSE, C, C 10/06/1913 0.70 CW, W, W, NW 12/03/1919 -0.25 CW,CSW, CW, NE 12/06/1919 -0.55 SW, A, SW, C 03/11/1921 -1.22 W, ANW, A, CW 19/09/1922 -0.43 W, CNW, AW, CW 02/06/1924 -0.21 C, CE, C, NW 21/07/1924 -1.12, -0.67, -0.12, 0.44 NW, NW, -, ANE 18/07/1926 -0.10, 0.18, 0.19, 0.02 A, A, A, SE 06/09/1927 -0.59 A, S, CSE, C 28/03/1930 -0.70 W, A, A, SW 15/11/1929 2.15 W, CNW, C, S 14/06/1931 -0.40 SW, A, AS, C 25/06/1935 0.05 S, A, NE, E 14/12/1936 2.37 SW, CW, SW, C 31/07/1939 -1.15, -1.72, -1.48, -1.05 S, CSW, C, CW 04/11/1940 1.30 SW, SW, CW, C 18/03/1947 -1.28 S, C, AS, CSE 11/01/1948 1.53 NW, CS, S, C 26/07/1957 0.16, -0.19, -0.62, -1.16 NW, ANW, SW, CW 05/08/1957 1.25, 1.11, 1.11, 1.25 A, A, SE, SE, 25/09/1957 -0.68 SE, CE, SE, E 12/12/1964 -1.24 W, SW, ASW, W 05/08/1973 -0.01, -0.17, -0.95, -1.81 C, CSW, C, C 15/08/1977 1.52, 1.67, 2.14, 2.54 A, ASE, SE, SE 11/05/1979 -0.01 C, A, S, CW 05/12/1979 2.07 W, SW, SW, W 30/12/1986 3.42 N, W, W, C 18/10/1987 -0.80 C, C, SW, CS

06/03/1998 1.24 C, C, W, CS 30/10/2000 2.26 W, C, CSW, C 08/06/2012 -1.16 S, C, C, C

Table 6

Statistically significant correlations (p = < 0.05) between November-February rainfall, average and maximum gauged daily flows at Cwmystwyth (1984-2010). NS = not significant

Jan NAO Feb NAO Nov NAO Dec NAO

Jan P 0.642 - - -

Jan ave. Q 0.644 - - -

Jan max. Q NS - - -

Feb P - 0.586 - -

Feb ave. Q - 0.599 - -

Feb max. Q - NS - -

Nov P - - 0.450 -

Nov ave. Q - - 0.443 -

Nov max. Q - - NS -

Dec P - - - 0.577

Dec ave. Q - - - 0.713

Dec max. Q - - - 0.547